← All news

AI News — July 11, 2026

#meta #agentic #pricing

Meta started charging for its own model for the first time: Muse Spark 1.1, shipped July 9 through the new paid Meta Model API, is an agentic coding model that tops the MCP Atlas tool-use benchmark (88.1) while pricing at $1.25/$4.25 per million tokens — roughly a quarter of Opus 4.8 and GPT-5.5 — marking Meta's turn from open-weight Llama toward a metered, agent-first product.

Model releases

  • [2026-07-09] Meta — Muse Spark 1.1 launches on the paid Meta Model API — Meta’s first time charging for its own model. Meta Superintelligence Labs shipped Muse Spark 1.1, a multimodal reasoning model built for agentic and coding work, and — for the first time — is charging developers to call Meta’s own model through the new Meta Model API (US developer public preview, $20 in free credits). It carries a self-managed 1M-token context window and native primary-agent/subagent orchestration (it can plan and delegate across parallel subagents, or run as a subagent that escalates back), with MCP and custom-skill support and reported zero-shot generalization to new tools. On tool-use benchmarks Meta claims the lead — MCP Atlas 88.1 (scaled tool use) and JobBench 54.7 (professional tool use) — but on coding it trails the flagships: SWE-Bench Pro 61.5 (Opus 4.8 69.2), Terminal-Bench 2.1 80.0 (GPT-5.5 83.4, Opus 4.8 82.7), DeepSWE 1.1 53.3 (GPT-5.5 67.0, Opus 4.8 59.0). Pricing is $1.25 / $4.25 per million input/output tokens — roughly a quarter of Opus 4.8 ($5/$25) and GPT-5.5 (~$5/$30). It matters because it is a strategic reversal: Meta’s model story was open-weight Llama you ran yourself, and Muse Spark 1.1 is a metered, agent-first API you rent — the third price-aggressive agentic model in a week after Grok 4.5 (July 8), but the first that puts a major open-weights lab behind a per-token meter. (official, source, enterprise)

    Muse Spark 1.1’s benchmark split is the useful signal for sizing it: it leads on tool-use (MCP Atlas 88.1, JobBench 54.7) but trails Opus 4.8 and GPT-5.5 across the SWE-Bench Pro and DeepSWE coding suites, so it reads as a model built to run an agent loop and call tools rather than to write the hardest code itself.

    For Solution Architects: In a multi-model agent system, that profile makes Muse Spark 1.1 a candidate for the orchestrator seat — planning and delegating across parallel subagents over MCP, where its 88.1 MCP Atlas score is the relevant strength — while you route the coding-heavy subtasks that need the higher SWE-Bench number to a pricier flagship, paying roughly a quarter of Opus 4.8’s rate for the plumbing.

Watch list

  • Gemini 3.5 Pro (GA) — entering the second week of July still a limited Vertex AI preview: no model card, published benchmarks, or pricing for the promised 2M-token context and Deep Think mode. Reporting continues to point to a July 17 target tied to a full architectural rebuild that scrapped the 2.5 Pro base, with early-tester complaints about token efficiency and coding cited as the reason for the slip; Google has confirmed none of it. With Grok 4.5 and Muse Spark 1.1 both now shipped, Gemini 3.5 Pro is the only major frontier flagship of the summer still entirely unlaunched. Watch for whether even a model card lands before the 17th. (unconfirmed) (prior coverage)

    With every rival now shipped and priced, the vacuum increasingly sets the bar for Gemini rather than the reverse — a July 17 launch that merely matches Grok 4.5 and Muse Spark on price would land as catch-up, so the scrapped-and-rebuilt architecture has to show up as a capability the others don’t have to justify the wait.

  • GPT-5.6 first independent benchmarks — with GA rollout reaching ChatGPT Plus around July 10, Sol/Terra/Luna are now in wide use, but the only public performance numbers remain OpenAI’s own. Muse Spark 1.1 just added another set of independently comparable coding scores (SWE-Bench Pro, Terminal-Bench, DeepSWE); watch for the first third-party eval that runs those same suites against Sol. (prior coverage)

    Because those three suites now carry published competitor numbers, the first independent run won’t just measure Sol — it will immediately rank it against Muse Spark, Opus 4.8, and Grok on identical coding tests, the summer flagships’ first head-to-head on neutral ground.

  • Fable 5 steady-state cost (July 12) — Anthropic’s Fable 5 stays included at up to 50% of weekly limits through 11:59pm PT July 12, after which access moves to metered usage credits at $10/$50 per million tokens. Muse Spark 1.1 at $1.25/$4.25 and Grok 4.5 at $2/$6 sharpen the question for teams that rewired production onto Fable 5: which paths actually need the top tier before the meter starts. Watch the cutover and whether standard Enterprise seats can enable usage credits at all. (official, prior coverage)

    This is the last briefing before the cutover itself — after tomorrow night the free window is gone, so whatever path-by-path triage teams haven’t finished by now happens against a live meter instead of ahead of it.

  • MCP spec finalization (July 28) — the 2026-07-28 release candidate (stateless core, Extensions framework, Tasks, MCP Apps, authorization hardening) is in its validation window with the Ruby/TypeScript/Python SDKs updating against it. Security analysts continue to flag the new Mcp-Method/Mcp-Name HTTP headers as fresh attack surface (protocol-confusion/desync, header data-leakage) to test before the cutover. Watch for the remaining SDK support landing. (official, prior coverage)